Optimal allocation of tradable emission permits under upstream-downstream strategic interaction

HAL Id: hal-00437645

https://hal.archives-ouvertes.fr/hal-00437645

Preprint submitted on 1 Dec 2009

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de

Optimal allocation of tradable emission permits under upstream-downstream strategic interaction

Joana Resende, Maria Eugénia Sanin

To cite this version:

Joana Resende, Maria Eugénia Sanin. Optimal allocation of tradable emission permits under

upstream-downstream strategic interaction. 2009. �hal-00437645�

OPTIMAL ALLOCATION OF TRADABLE EMISSION PERMITS UNDER UPSTREAM-DOWNSTREAM STRATEGIC

INTERACTION

Joana RESENDE María Eugenia SANIN

November 2009

Cahier n° 2009-50

ECOLE POLYTECHNIQUE

CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE

DEPARTEMENT D'ECONOMIE

Route de Saclay 91128 PALAISEAU CEDEX

(33) 1 69333033

http://www.enseignement.polytechnique.fr/economie/

mailto:chantal.poujouly@polytechnique.edu

Optimal allocation of tradable emission permits under upstream-downstream strategic interaction

Joana Resende

CORE, Université catholique de Louvain CETE,University of Porto E-mail: jresende@fep.up.pt

and

María Eugenia Sanin

Département d’Économie de l’École Polytechnique CORE, Université catholique de Louvain

Chair Lhoist Berghmans in Environmental Economics and Management E-mail: Maria-Eugenia.Sanin@uclouvain.be

Version: November 2009

In this paper we account for the fact that Cournot equilibrium strategies in the sector under environ- mental regulation depend on …rms’interaction in the permits market (and vice versa). In this context, we show that the cost-e¤ective allocation of permits between …rms must compensate the cost-rising strategies exercised by the stronger …rm (in the output market). Then, taking into account the previous result, we use a simulation to obtain the optimal allocation of permits between …rms as a function of output market characteristics, in particular as a function of goods substitutability that serves as an indicator for the de- gree of price competition. The simulation allows us to determine how output market characteristics a¤ect di¤erently optimal permit allocation depending on the regulator’s objective.

ACKNOWLEDGMENTS

We wish to thank Paul Belle‡amme, So…a Castro, Thierry Bréchet, Jean J. Gabszewicz and Juan Pablo Montero for their stimulating comments. The usual disclaim applies.

1. INTRODUCTION

In this paper we investigate how strategic interaction in an olygopolistic output market may

a¤ect the e¤ectiveness of environmental regulation based on tradable emissions permits. With this

purpose, we assume that two Cournot producers of a (di¤erenciated) polluting good must comply

with environmental regulation by holding a tradable emission permit per unit of (non-abated)

pollution generated by goods’ production. This makes tradable emissions permits an essencial

input. We model the interaction between the tradable emissions permits market (upstream) and the output market (downstream) by considering the following three-stages game: in the …rst stage, one of the …rms sets the price of permits alone; in the second stage permits are traded; and,

…nally, in the third stage, …rms compete à la Cournot in the (di¤erenciated) output market. Our model follows Montero (2002) and the traditional theory on upstream-downstream interaction (from Salinger, 1988 to Ordover, Salop and Saloner, 1990), assuming that the upstream (permits) market is cleared …rst. However, di¤erently from the latter, our model captures the nature of tradable emissions permits markets by allowing …rms’position in the upstream market to be endogenous.

Our assumption regarding the fact that a …rm has a …rst-mover advantage over the other in the permits market captures the fact that most permits markets created so far have been organized in subsequent "phases": in a …rst phase, only the most polluting …rms receive tradable permits that can be traded in the permits market, whereas, their less polluting competitors are only included in a second phase (usually two to four years latter). This was the case in the EU Emission Trading Scheme (EU-ETS), in the US Acid Rain Market and in the NOx budget program, among many others. In this respect, Boemare and Quirion (2002) note that EU-ETS coverage is con…ned to a limited number of sectors, and that there is no provision for voluntary "opt in" by …rms below the threshold size that assign …rms to the …rst phase. They argue that such "opt in" provisions could help dilute any emerging market power.

The risk of market power in a tradable permits market covering more than one sector, like the EU-ETS, seems smaller than in the case where only one olygopolistic sector is included

or in the case that one olygopolistic sector represents a big portion of the total permits market.

According to Kolstad and Wolak (2008), the olygopolistic …rms participating in the Californian electricity market (CAISO) behaved strategically in the Los Angeles market for NOx emissions called RECLAIM (Regional Clean Air Incentives Market) to enhace their ability to exercise market power in the CAISO. Such …rms, shown to exhert unilateral market power in the CAISO by Wolak (2003), were allocated 56% of total initial stock of permits. In the same line, Chen et al. (2006) compute equilibrium behavior considering the interaction between the NOx budget program and the Pennsylvania-New Jersey-Maryland (PJM) electricity market. Due to the high concentration of the PJM market

, six large electricity generators alone account for 90% of emissions in the referred permits market. Chen et al. (2006) …nd that a Stackelberg leader with a long position in the permits market could gain substantial pro…ts by withholding permits and driving up permits costs for rival producers.

The supporters of environmental regulation based on tradable emissions permits argue that the

1In this respect Kolstad and Wolak (2008) argue that "although big fossil-fuel generation units produce a fraction of greenhouse gas emissions (GHG), they are likely to be early and substantial participants in any GHG emission permits trading scheme". In the same line, Linares et al. (2006) argue that electricity generators represent more than 50% of emissions covered by the EU-ETS market.

2Its Her…ndahl-Hirschman Index was 0.154.

creation of an emission permits market allows to reach the pollution reduction target (generally called "the cap") in a cost-e¤ective manner (Montgomery, 1972) and with a minimum information cost for the regulator (namely, concerning information regarding pollution abatement technologies).

This argument has been challenged, …rst, by Hahn’s (1985) dominant-fringe model which argues that the existence of market power reduces the cost-e¤ectiveness of tradable emissions permits because a dominant …rm manipulates the price of permits to reduce its own emission abatement costs.

Misiolek and Elder (1989), inspired in Salop and Sche¤man (1987), and Eshel (2005), also relying on a dominant-fringe setting, show that technological linkages between permits and output markets would give raise to rival’s cost-rising strategies by the dominant …rm, which would introduce an additional type of market distortion. Fehr (1993) and Sartzetakis (1997) have also challenged the e¤ectiveness of environmental regulation based on tradable emissions permits, showing that, in a context of strategic interaction, emission permits markets could lead to monopolization or excessive entry barriers. Although these two last papers have considered strategic interaction in the output market, their objective is just to focus on the previously mentioned corner solutions instead of assessing the e¤ects of strategic interaction on permits prices (or optimal permits allocation). In fact, Fehr (1993) assumes that downstream …rms buy permits for a given supply, whereas Sartzetakis (1997) assumes a competitive permits market.

The empirical papers’…ndings mentioned earlier in this introduction, together with the recent discussion on "windfall" pro…ts in oligopolies

subject to environmental regulation based on tradable emissions permits, suggest the need to account for the way this regulation a¤ects pro…ts and …rm’s behaviour when the downstream market is olygopolistic. This paper wishes to contribute to the literature on non-competitive emissions trading by accounting for an olygopolistic downstream market and discussing how its characteristics determine the optimal permits allocation rule. In this way, we are able to identify three di¤erent channels of market distortion. The …rst market distortion is due to market power in the permits market. Everything else the same, as in Hahn (1985), the …rst- mover in the permits market manipulates the price of permits to reduce its own environmental costs.

The other market distortions are related to …rms’market power in the output market, which gives both …rms the incentives to underinvest in pollution abatement. By using permits for production,

…rms become more competitive in the output market as they reduce their own marginal abatement costs, while increasing their rival’s marginal abatement costs. Di¤erently from the dominant-fringe literature’s …ndings, we show how strategic interaction in the output market endows the follower in the permits market to adopt rival’s cost-rising strategies as well. In this context, we show how the possibility of simultaneous adoption of rival’s cost-rising strategies may either aggravate of ameliorate the cost-e¤ectiveness of environmental regulation based on tradable emissions permits markets

. The net impact of the simultaneous adoption of rival’s cost-rising strategies depends on

3The Arcelor-Mittal case in April 2009 is one example from the steel industry under the EU-ETS.

4It is worth noting that the conclusions driven in this paper are only concerned with tradable emissions permits

…rms’position in the permits market (buyer or seller) as well as on the interplay of …rms’actions in the output market.

In this line, we propose an optimal criterion to allocate permits between …rms. We argue that the speci…c features of this criterion are highly sensitive to (i) the regulator’s objective (cost- e¤ectiveness of pollution abatement, maximization of …rms pro…ts, maximization of social wel- fare,...); and (ii) the speci…c characteristics of the market, in particular the degree of substitution between goods. The latter point underlines how, even if tradable emissions permits may lead to cost-e¤ectiveness of pollution abatement (or any of the mentioned objetives), when the downstream market is olygopolistic this cannot be done at the minnimum information cost.

2. BASIC SETUP

Consider a duopolistic market, in which …rms (…rm i and …rm j) compete in quantities producing imperfect substitute goods i and j. Quantity y

represents the production of good k = i; j, p

represents its price and c

[y

] its production cost, with c

[y

] > 0 and c

[y

] > 0:

The production of goods i and j generates polluting emissions as a by-product. The parameter

> 0 represents output polluting intensity. Firms are subject to environmental regulation based on a cap and trade system. Under such environmental regulation, each …rm must hold a number of permits E

equal to the amount of pollution y

emitted during production of y

: Then, this regulation creates a scarce input, tradable permits, that are available up to the cap S: Then, the total stock of permits S is allocated for free among …rms. A percentage is received by …rm i and a percentage (1 ) by …rm j. The percentage 2 [0; 1] and the cap S are chosen exogenously by the regulator, according to the pollution control target. In a framework where only polluting …rms trade in rights, the regulator’s decisions regarding the allocation of permits between …rms and the decision regarding the total cap on emissions can be analyzed independently (Eshel, 2005). In this paper we restrict attention only to the allocation decision

.

When the permits received for free, S and (1 ) S respectively, do not coincide with the emissions provoked by production of the optimal quantity of output y

, …rms choose either to

regulation and cannot be extrapolated to other environmental regulation instruments like taxes or standards.

5The cap on pollution is generally …xed by the regulatory authority with the help of experts -like the IPCC 1990 Scienti…c Assessment in the case of the Kyoto protocol and its european side agreement for the creation of the EU-ETS- that state the impacts of pollution and the pollution reduction required to diminish those impacts to an acceptable level. By creating property rights for an amount equal to the acceptable environmental damage, the authority ensures that the marginal cost of pollution reduction (i.e. the property right’s price) equals the marginal bene…t of such reduction (i.e. the unit improvement in environmental quality).

abate some of these emissions or to engage in permits trading. That is,

E

= S x

0; (1)

E

= (1 )S x

0; (2)

a

= y

E

0; k = i; j; (3)

where x

denotes the amount of permits sold (when x

> 0) or bought (when x

< 0) by …rm k;

and a

stands for the level of emissions abated. Abatement of polluting emissions is costly. To abate a

polluting emissions, …rm k has a cost of h

[a

] ; with h

[a

] > 0 and h

[a

] > 0:

In this regulatory framework, …rms interact in two technologically-linked markets: the permits market (upstream market) and the output market (downstream market). In the output market, both …rms exert some degree of market power since …rms take their quantity decisions simultane- ously. In the permits market, the degree of market power is asymmetric, with one of the …rms (say

…rm j) having a …rst-mover advantage.

We model interaction in these technologically linked markets using a sequential game theoretical approach. The players are the two …rms, the payo¤s are …rms’ pro…ts and strategies correspond to the vector q; E

(q) ; y

(q; E

) ; y

(q; E

) ; with q 2 R

; E

2 R

; and y

2 R

. The timing of the game is the following: in the …rst stage, …rm j sets alone the price of permits (q); in the second stage, …rm i observes the posted price of permits (q) and chooses the amount of permits to use for production (E

) ; which determines the amount of permits to buy or sell (x

). Firm j clears the permits market (x

= x

) : Finally, in the third stage, given optimal use of permits (E

) and the corresponding optimal permits trade (x

), …rms simultaneously interact in the output market, strategically competing on quantities (y

and y

).

3. CHARACTERIZATION OF THE EQUILIBRIUM

We rely on the notion of subgame perfect Nash equilibrium (SPNE) to investigate …rms’optimal behavior. The optimization problem of …rms in each stage is detailed in Apendix A.

Definition 1. The SPNE corresponds to the vector of strategies q ; E

(q) ; y

(q; E

) ; y

(q; E

)

such that

(i) Given q and E

; y

is the best response of …rm i to y

and vice-versa;

(ii) Given y

(q; E

) ; E

(q) is the best response of …rm i to q; and

(iii) Given y

(q; E

) and E

(q), q maximizes total pro…ts of …rm j:

We consider that goods are imperfect substitutes and inverse demand functions are such that:

@p

[y

; y

]

@y

< @p

[y

; y

]

@p

< 0: (4)

According to the previous de…nition, in the last stage …rms simultaneously choose the output levels that solve problems (17) and (18), respectively. When considering linear

demands, the in- terior

equilibrium vector y

(E

; q) ; y

(E

; q) is directly obtained from the system of …rst order conditions (FOCs

), i.e.:

p

[y

; y

] +

i

y

= c

[y

] + h

[a

] p

[y

; y

] +

j

y

= c

[y

] + h

[a

] : (5)

Given outcomes in the permits market and rival’s output choice, in equilibrium each …rm chooses the equilibrium production level y

(E

; q) for which there is a perfect balance between the marginal revenue and the marginal cost (including abatement marginal costs).

According to equilibrium conditions (5), outcomes in the permits markets a¤ect output decisions via marginal abatement costs. The following lemma summarizes this transmission mechanism.

Lemma 1. The larger the amount of permits used for production by …rm k; the larger its equi- librium output level and the lower the equilibrium output level for its rival k. It follows:

@y

[E

]

@E

> 0; (6)

@y

[E

]

@E

< 0:

Proof. See the Appendix.

The result in (6) is due to the fact that, the larger the amount of permits used for production by

…rm k, the lower its abatement needs a

, the lower its marginal abatement costs h

(a

); and there- fore, the higher its output production. This direct e¤ect is reinforced by strategic substitutability

6By considering linear demand functions we ensure that second order e¤ects are zero, i.e. for each …rmkit is the case that ^@

2p_k[y_k;y _k]

@y²_k =^@

2p_k[y_k;y _k]

@y_k@y _k = ^@

2p_k[y_k;y _k]

@y²_k = 0:

7Essentiatlly, we restrict our attention to the parameters for which the marginal revenue is higher than the marginal cost when the strategic variable is equal to zero (q = 0; Ek = 0, yk = 0). When this is not the case, equilibrium outcomes may correspond to corner solutions, in which …rms’optimal behavior may di¤er from the …rst order conditions derived in this paper. A su¢ cient condition to guarantee that we are in an interior solution is to assume that, wheny_k= 0;it is the case thatp_k[yi; yj]> c⁰_k[y_k] + h⁰_k[y_k]8y _k, independently of the rival’s output decision.

8Second order conditions are ful…lled as long as: ^@

2p_k[y_i;y_j]

@y²_k yk 2^@pk_@y^[yi;yj]

k +c⁰_k[yk] + ⁰_k[ak]:In the case of linear demands and strictly convex cost functions (production costs and abatement costs), this condition is always met.

of …rms’ output decisions. Accordingly, the result in (6) is independent of which of the two rival

…rms is the most e¢ cient in terms of abatement.

In the second stage, …rm i chooses the amount of permits to use for production (E

) after observing the price of permits (q). When deciding E

; …rm i anticipates the strategic interaction that will take place in the output market. Therefore, when deciding E

, …rm i takes into account the marginal pro…tability of permits transactions (q h

[a

]) as well as the impact of E

in the output market outcome. In the interior equilibrium, E

[q] is obtained from the …rst order condition

to problem (15), that is,

dp

[E

] dE

y

[E

] +

[E

] @y

@E

= q h

[y

[E

]]; (7)

with

[E

] = p

[E

] c

[y

[E

]] h

[y

[E

]] and

i

=

i

@y_i[Ei]

@Ei

+

j

@y_j[Ei]

@Ei

:

In equilibrium, E

[q] is such that …rm i does not have any opportunity to increase its total pro…ts by trading-o¤ output pro…ts by pro…ts due to permits’ transactions. More precisely, in equilibrium, any variation in the pro…ts from permits transactions caused by a marginal variation of E

(equal to q h

, in the RHS of (7)) is perfectly o¤set by an equivalent variation in output pro…ts (which is given by the LHS of (7)).

Lemma 2. For a given price of permits q, …rm i always abates less than e¢ ciently. The di¤ er- ence between permits price and marginal abatement costs is given by:

q h

[E

] = @p

@y

@y

[E

]

@E

y

[E

] > 0: (8)

Proof. See the Appendix.

The lemma shows that, in a scenario of strategic interaction in the output market, abatement decisions of …rm i are such that q > h

[E

]: Accordingly, in equilibrium, …rm i always abates less than what a competitive …rm would abate. Firm i strategically chooses to forego pro…ts from permits transactions to bene…t from a better position in relation to its rival in the output market.

Due to strategic interaction in the output market, despite being a price-taker in the permits market,

…rm i is able to use its decisions concerning E

to reduce its overall marginal costs, while increasing its rival marginal costs (since the rival is responsible for clearing the permits market). Firm i is then able to increase its market share in the output market, obtaining higher output pro…ts than the pro…t level corresponding to the choice of E

that leads to q = h

.

9Our analysis is valid when the second order condition ^{@2 i[Ei;q]}

@E_i² <0holds, i.e. when:

2@y_i

@Ei

dp_i dEi

< c⁰⁰_i @y_i

@Ei 2

+ 1 @y_i

@Ei 2

h⁰⁰_i:

Finally, in the …rst stage, …rm j quotes the price q that solves the following …rst order condition:

q x

[q]

@E_i[q]

@q

h

= dp

dE

y

+

[E

[q]] @y

@E

: (9)

where we still focus exclusively on interior solutions.

The equilibrium price of permits q that satis…es (9) guarantees that …rm j is exploiting all existing pro…t opportunities (considering the permits market as well as the output market). In equilibrium, variations in output pro…ts induced by marginal variations in q (given by the RHS in condition (9)) are exactly compensated by variations in the pro…ts associated with permits’

transactions (given by the LHS in condition (9)).

Lemma 3. Everything else the same, strategic competition in output quantities leads to a permits price that may be either higher or lower than the e¢ cient one q b = h

. In equilibrium, the di¤ erence between the price of permits and the marginal abatement cost of …rm j is equal to:

q h

= x

[q]

@E_i[q]

@q

@p

[y

; y

]

@y

@y

@E

y

[E

[q]]: (10)

Proof. See the Appendix.

According to (10), we could observe a price of permits q equal to h

only by chance. The di¤erence between q and h

can be decomposed in two e¤ects. The …rst contribution to the di¤erence between permits price and marginal abatement costs is due to the market power of …rm j in the permits market (this corresponds to

i[q]

@q

in condition (10)). Focusing exclusively on this e¤ect, condition (10) implies that, everything else the same, when …rm j is a net-seller of permits (x

< 0), it enjoys a positive mark-up over h

as a result of its …rst mover advantage in the permits market. In contrast, when …rm j is a net-buyer of permits (x

> 0) ; …rm j will mark down permits price (price discount).

Besides the market power exerted by …rm j in the permits market, there is an additional e¤ect that contributes to the di¤erence between price of permits q and its marginal abatement costs h

: This additional source of ine¢ ciency corresponds to the positive term

i

@y_i

@Ei

y

> 0 in

condition (10). Independently of …rms’position in the permits market, the latter e¤ect derives from

the technological link that exists between the permits market and the output market and is always

positive. All the rest being equal, …rm j has incentives to make permits more expensive for its rival

i, reducing …rm i

s output production, while increasing its own market share in the output market.

4. COST-EFFECTIVENESS OF POLLUTION ABATEMENT

From Lemma 2 and Lemma 3, it follows that market mechanisms based on tradable emissions permits may not entail an e¢ cient allocation of abatement e¤orts between …rms.

This result is in line with previous literature. In a dominant-fringe setting, Hahn (1985) shows how market power in the permits market could damages cost-e¤ectiveness in pollution abatement.

Also using a dominant-fringe setting, Misiolek and Elder (1989) and Eshel (2005), among others, have suggested that the cost-e¤ectiveness in pollution abatement could also be a¤ected by dominant

…rms’ability to use emission permits to raise rivals’costs. These two e¤ects correspond to the e¤ects described in Lemma 3.

However, in the context of our model, even when …rm i is a price taker in the permits market, strategic interaction in the output markets creates a new e¤ect, which a priori might either aggravate or ameliorate the cost-e¤ectiveness of abatement e¢ ciency. This e¤ect stems from the fact that we observe cost-rising strategies also in the case of …rm i (see Lemma 2). More precisely, in equilibrium, we obtain that

h

[a

] h

[a

] = x

[q]

@E_i[q]

@q

+

S0

@p

@y

@y

@E

y

>0

+ @p

@y

@y

@E

y

<0

: (11)

According to (11), the equilibrium di¤erential between …rms’ abatement costs can be decom- posed in three e¤ects. The …rst e¤ect is given by

@q

and it stems from …rm j

s market power in the permits market (Hahn, 1985). The second e¤ect is given by

i

@y_i

@Ei

y

and it stems from

…rm j

s ability to use its decision regarding the price of permits to raise …rm i

s marginal costs.

This e¤ect is equivalent to the output market e¤ect emphasized in the dominant-fringe literature, since …rm j is also strategic in the output market. Finally, the third e¤ect is given by

j

@y_j

@Ei

y

and stems from …rm i’s ability to use its decision regarding permits’use to raise …rm j

s marginal costs, i.e. to compensate the previous e¤ects. This third e¤ect due to strategic interaction in the output market was not considered in previous literature, like Misiolek and Elder (1989) or Eshel (2005). Whether this third e¤ect aggravates or ameliorates cost-e¤ectiveness of abatement depends on …rms’position in the permits market, as well as on the interplay of …rms in the permits market and in the output market.

When …rm j is a net-seller of permits, permits are always over-priced

(q > h

). Ceteris paribus, the higher price of permits induces a reduction in permits demanded by …rm i due to a

1 0When …rmj is a net-seller of permits, permits are over-priced for two reasons: according to (10), on the one hand, there is the output market e¤ect corresponding to ^@pj_@y^[yi;yj]

i

@y_i

@E_iyj >0and, on the other hand, there is an addicional mark-up of ( ^{S E}i[q])

@Ei[q]

@q

due to …rm i’s market power in the permits market (…rst mover advantage).

move along …rm i

s demand of permits. But …rm i is also strategic. As described in Lemma 2,

…rm i

s strategic behavior shifts permits’demand upwards, o¤setting (at least partially) the market distortion associated with the market power of …rm j: However, when …rm i

s strategic in‡uence in this market is very strong (i.e.

j

@y_j

@Ei

y

is very large), this e¤ect may yield instead market distortions with an opposite nature, i.e. with …rm i under-investing in pollution abatement (such that h

[a

] < h

[a

])

even more than …rm j.

The previous analysis remains valid when …rm j is a net buyer of permits but …rm j

s market power in the permits market is relatively limited and its incentives to adopt a raising rival’s costs strategy more than compensate the incentives to mark-down permits price, i.e.

i[q]

@q

@pj

@yi

@y_i

@Ei

y

>

0:

When …rm j is a net buyer of permits and it exerts a substantial degree of market power in the permits markets, permits prices are under-priced, i.e. the …rst term in the RHS of (10) is higher in absolute value that the second term.

Ceteris paribus, this yields a downward move along …rm i’s supply of permits. In this case, …rm i

s strategic e¤ect aggravates the market distortion provoked by …rm j by shifting supply of permits upward, which makes the di¤erence in (11) larger.

To the light of the previous results, we conclude that market mechanisms alone may not (and in general do not) lead to cost-e¤ectiveness of pollution abatement. The extent to which …rms’deviate from the cost-e¤ective pollution abatement solution depends on …rms’initial permits’endowments, together with the incentives for rising rival’s costs. Therefore, for a given cap on permits (S) ; the regulator may use the allocation of permits among …rms ( ) as a policy instrument to reestablish the cost-e¢ cient result or to promote welfare-enhancing policies. Such choice depends on the regulator’s objective. The policy implications of our results are analyzed with further detail in the next section.

5. REGULATION AND POLICY IMPLICATIONS

This section investigates how the environmental regulator may rely on his decision regarding

…rms’ initial endowment of permits to restablish cost-e¤ectiveness in the allocation of abatement e¤orts between …rms. Afterwards, we consider a broader regulatory objective, investigating the impact of permits allocation rules on total social welfare.

From Lemma 2, we conclude that strategic interaction in the output market results in an upward shift of …rm i

s demand (or supply) of permits, which allows …rm i to under-invest in pollution abatement, for any given price of permits. Depending on the interplay between the market power in the permits market and the incentive to adopt a rival’s cost-raising strategy, …rm j might be interested in under-invest or over-invest in pollution abatement.

1 1From a theoretical point of view, such situation may occur. Nevertheless, there should be a considerable degree of asymmetry between …rms so that …rmi⁰s strategic e¤ect in the output market more than compensates …rmj⁰s strategic e¤ect in the output market as well as its market power in the permits market.

Proposition 1. A regulator is able to use the permits’allocation rule to restore cost-e¤ ectiveness of pollution abatement. The cost e¤ ective allocation rule

is implicitly given by the value of

which solves:

@p

@y

@y

@E

y

+ @p

@y

@y

@E

y

= S

E

[q]

@E_i[q]

@q

: (12)

The sign of the LHS of (12) is a priori undetermined. If the e¤ect of …rm i

s cost-raising strategy dominates the e¤ect of …rm j

s cost-raising strategy, i.e.

j

@y_j

@Ei

y

>

i

@y_i

@Ei

y

; when the cost-e¤ective allocation rule is applied by the regulator, …rm j must end up being a net seller of permits to eliminate …rm i

s incentive to overuse emission permits in production. In contrast, when

j

@y_j

@Ei

y

<

i

@y_i

@Ei

y

; the LHS of (12) is negative, and therefore, the e¤ect of …rm j

s cost-raising strategy by itself is more than enough to compensate the e¤ect inherent to …rm i

s cost-raising strategy. To avoid under-investment in pollution abatement by …rm j; the regulator must allocate permits in such a way that …rm j becomes a net-buyer of permits so that its market power in the permits market partially o¤sets its cost-raising strategy in the output market.

However, the cost-e¤ectiveness of pollution abatement may not be the only aspect that the regulator considers when deciding permits’allocation rules. In fact, given the technological linkages between permits and output markets, we observe that outcomes in the output market are also a¤ected by the regulator’s choices with respect to the permits allocation.

In particular, along the equilibrium path, the output level of the …rm receiving more permits increases while the output of the …rm receiving less permits decreases:

@y

@ > 0 and @y

@ < 0 (13)

The mechanism behind this result is the following: …rst, an increases of leads to a decreases in permits’price. This reduction of permits price entails a downward move along …rm i

s demand of permits (or supply of permits, if …rm i is a net-seller of permits), which entails an increase in the use of permits for production by …rm i: As a consequence, …rm i

s marginal abatement costs are lower than before, yielding an increase in …rm i

s output production. The opposite occurs to

…rm j; that looses market share after an increase of . The impact on total output and consumers’

welfare depends on whether the increase of y

more than compensates the decrease of y

after the variation of :

The cost-e¤ective permits allocation rule (

) fails to internalize this type of considerations. In

fact, optimal allocation rules may change considerably depending on the regulator’s objective. Some

examples of possible objectives to be pursued by the regulator are: abatement cost-e¤ectiveness,

maximization of joint pro…ts, maximization of consumers’surplus (even with some preference over

some type of consumers), maximization of total welfare, or other. Depending on the scope of the

regulator’s activity and the speci…c objective pursued by the regulator, there may be contradictory

recommendations regarding the optimal allocation rule :

In the following section we introduce an example that illustrates the previous point. In partic- ular, the following section shows how output demand characteristics in‡uences optimal allocation of permits between …rms.

6. THE IMPORTANCE OF OUTPUT DEMAND CHARACTERISTICS

To illustrate how sensitive optimal allocation rules may be both in relation to the speci…c characteristics of the industry under regulation and to the objective pursued by the regulator, consider an industry with the following characteristics: (i) the inverse demand for good k is given by p

[y

; y

] = 25 + "

2y

y

; with 0 < < 2 and "

= 0:1 if k = j; (ii) …rms’production technology is similar, with c

[y

] =

; (iii) the intensity of pollution is equal to = 0:8; (iv) the total stock of emission permits is exogenously …xed by the regulator to meet the pollution control target and we consider it to be S = 3; which constrains …rms’ production plans. Finally, (v) …rms’abatement technologies are given by h

[a

] =

and h

[a

] =

: From (v) follows that the …rst-mover in the permits market (…rm j) owns a more e¢ cient abatement technology, even if the degree of asymmetry between …rms’ abatement technologies is rather limited. This is in line with Hahn (1985) and Eshel (2005), among others, that de…ne the dominant …rm in their dominant-fringe setting as the …rm with the lowest marginal abatement cost function.

In this context, we compute the equilibrium outcomes as in Section 3 but now considering the speci…c functions detailed in points (i) to (v). We then illustrate the degree of output substitutabil- ity, as a measure of competition between …rms in the output market, as well as the objective pursued by the regulator, a¤ect optimal allocation rules. Regarding, the impact of output substitutability, we compute optimal allocation rules for progressively higher degrees of substitutability between goods, comparing outcomes for = 0:01; = 0:4; = 1:5; and = 1:99. Concerning the impact of the objective pursued by the regulator, we compare optimal allocation rules under four di¤erent regulatory objectives: (i) cost-e¤ectiveness of abatement e¤ort (CE); (ii) maximization of …rms’

joint pro…ts (J P ); and, …nally, (iii) maximization of total social welfare (W ). Regarding the way we compute consumer’s surplus to be able to consider it when calculating social welfare, we use the partial equilibrium analysis from Belle‡amme and Peitz (2010). Table 1 summarizes our results:

Reading each line separately, we conclude that the optimal allocation of permits varies according

to the regulator’s objective. When the regulator wants to promote cost-e¤ectiveness of pollution

abatement (CE), he voluntary chooses to ignore the impact of permits decisions on the output

market. Table 1 shows that, under a cost-e¤ective allocation rule, the regulator must allocate more

permits to the …rm that owns the less e¢ cient abatement technology. Since in this example we con-

sider h

< h

, the asymmetric allocation in favor of …rm i is needed to reduce its abatement needs

(in comparison to …rm j

s abatement needs) up to the point in which h

[a

] = h

[a

]: In relation to

Substitution CE JP W

= 0:01

= 54:1%

= 54%

= 69%

= 0:4

= 54:4%

= 53%

= 61%

= 1:5

= 54:9%

= 51%

= 60%

= 1:99

= 55:2%

= 51%

= 60%

TABLE 1

Optimal permits allocation as a function of substituibility according to regulatory objective

the way output market characteristics in‡uence the cost-e¤ective allocation rule, as sustituability decreases, rising-rival’s cost strategies are weaker (due to less competition) and therefore the lower must be the compensation (through a higher permits allocation) to the less e¢ cient …rm. However, the cost-e¤ective allocation rule is not very sensitive to changes in the degree of output substitua- bility. This is due to the fact that the latter only a¤ects the former indirectly, i.e. through the e¤ect that a change in output substitutability has on output production, and consequently in abatement needs.

Turning now to the maximization of joint pro…ts (JP), the optimal allocation of permits once more favours the least e¢ cient …rm i: This is the case because, when giving more permits to the least e¢ cient …rm, the increase in its output production (and its revenues) more than compensates the decrease in its rival’s production (and pro…ts). The changes in the pro…t-maximizing allocation as substituability changes are stronger than in the cost-e¤ective allocation. In particular, we observe that, as substituability decreases, …rm i’s increase in production (and pro…ts) after an increase in (keep in mind equation (13)) is accompaigned by a smaller decrease in j’s production (and pro…ts).

Then, the lower the substitution between goods, the higher the amount of permits that should be allocated to the less e¢ cient …rm.

Finally, the welfare maximizing (W) allocation rule gives even more permits to the least e¢ cient

…rm than the pro…t maximizing allocation rule. This is the case because the latter only accounts

for the fact that the increase in the least e¢ cient …rm’s output production after an increase in

more than compensates the decrease in the most e¢ cient …rm’s output production. Instead, the

welfare maximizing allocation rule additionally accounts for the fact that the mentioned changes

in quantities also a¤ect prices and consequently consumer’s surplus. In particular, the negative

variation in the least e¢ cient …rm’s price p

(thin line in Figure 1) more than compensates the

positive variation in the most e¢ cient …rm’s price p

: These price changes make consumers better-

o¤ as increases.

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

13.34

Permits allocation Output price

Permits allocation Output price

Figure 1: Output prices as a function of :

Concerning the impact of output substitutability on the welfare-maximizing allocation rule, we

…nd that as the substituability between goods decreases, the higher the gap between the negative variation of p

with an increase in and the positive variation of p

with an increase in : As a result, as substituability decreases, the higher it must be the fraction of permits allocated to …rm i to compensate the fact that p

is higher than p

(due to higher marginal abatement costs).

It is worth noting that the amount of permits allocated to the least e¢ cient …rm increases as we move rightwise in Table 1. This is because, when giving more permits to the least e¢ cient …rm, the cost-e¤ective allocation rule takes into account the bene…t in terms of cost minimization while the pro…t-maximizing allocation takes into account, additionally, the positive e¤ect on overall quantity produced and, …nally, the welfare-maximizing allocation additionally considering the bene…ts for consumers in terms of lower prices due to the latter increase in production.

The analysis regarding the regulatory possibilities as a function of goods substituability becomes particularly relevant when thinking of environmental policy that may a¤ect the redistribution of production between goods and, through that redistribution, may harm least favoured consumers.

7. CONCLUDING REMARKS

We have investigated how strategic interaction in the output market together with market power

in the permits market may a¤ect the e¤ectiveness of environmental regulation based on tradable

emissions permits. We propose a model of upstream-downstream competition, whose major features

are the following: (i) explicit consideration of the technological linkages between the permits market (upstream) and the output market (downstream); (ii) strategic interaction in the output market, with …rms competing à la Cournot; and (iii) market power in the permits market.

In line with previous literature dealing with the impact of market power on the e¤ectiveness of environmental regulation based on tradable emissions permits, we also conclude that emission per- mits markets may lead to outcomes that di¤er from the cost-e¤ective pollution abatement solution.

Besides market distortions associated with market power in the permits market (as in Hahn, 1985) and market distortions associated with the possibility of having a dominant …rm adopting rival’s cost-rising strategies (as in Misiolek and Elder, 1989 and Eshel, 2005), we show that strategic interaction in the output market may give raise to an additional source of market distortion. This market distortion is associated with the possibility that a follower in the permits market (exerting some degree of market power in the output market) may adopt a rival’s cost-rising strategy as well.

The net e¤ect of the three market distortions is a priori unknown. Strategic e¤ects associated with followers’cost-rising strategies might ameliorate or aggravate the cost-e¤ectiveness of emission permits markets but it may as well aggravate it, depending on the interplay of …rms’ actions in the permits and in the output market, as well as on …rms’position in the permits market (buyer/

seller of permits). Since …rms’ position in the permits market is directly related to …rms’ initial endowments of permits, to the light of the regulator’s speci…c objective, it is possible to design optimal allocation rules that restore cost-e¤ectiveness (or achieves another policy objective).

In relation to the design of optimal allocation rules, this paper highlights that the optimal

allocation rule may be extremely sensitive to the speci…c characteristics of the output market, in

particular to goods substituability. Then, in line with Sartzetakis (1997), this paper emphasizes

that the regulator is often faced with extremely demanding information needs.

REFERENCES

[1] Boemare, C. and Quirion, P., (2002), Implementing greenhouse gas trading in Europe: lessons from economic literature and international experiences, Ecological Economics, 43, 213-230.

[2] Chen, Y., Hobbs, B.F., Ley¤er, S. and Munson T.S., (2006), Leader-follower equilibria for electric power and NOx allowances markets, Computational Managment Science, 3, 307–330.

[3] D. M. D. Eshel, "Optimal allocation of tradable pollution rights and market structures", Jour- nal of Regulatory Economics, 28:2 205-223 (2005).

[4] Fehr, N.-H. M. v. d. (1993), "Tradable Emission Rights and Strategic Interaction". Environ- mental and Resource Economics 3. pp. 129-151.

[5] Hahn, R. (1985), Market power and transferable property rights. Quarterly Journal of Eco- nomics, 99. pp.753–765.

[6] IPCC (1990), Climate change: The IPCC Scienti…c Assessment.

[7] Kolstad, J. and F. Wolak, (2008), Using emission permits prices to rise electricity prices:

Evidence from the California electricity market, Harvard University, mimeo.

[8] Linares, P., Santos, J., Ventosa, M. and Lapiedra, L., (2006), Impacts of the European Emis- sions Trading Scheme Directive and Permit Assignment Methods on the Spanish Electricity Sector, The Energy Journal, 27(1), 79-98.

[9] Misiolek, W. S. and H. W. Elder (1989), Exclusionary manipulation of markets of pollution rights. Journal of Environmental Economics and Management, 16. pp. 156-166.

[10] Montgomery, D. W. (1972) Market in Licenses and E¢ cient Pollution Control Programs, Journal of Economic Theory, pp. 395-418.

[11] Ordover, J., Saloner, G. and Salop, S., (1990), Equilibrium Vertical Foreclosure, The American Economic Review, 80, pp. 127-142.

[12] Salinger, M. (1988) "Vertical Mergers and Market Foreclosure", The Quarterly Journal of Economics, , vol.103, 345-356.

[13] Salop, S.C. and D.T. Sche¤man (1983), Raising rivals’ costs, American Economic Review Papers and Proceedings 73, 267-271.

[14] Sartzetakis E.S., "Tradable Emission Permits Regulations in the Presence of Imperfectly Com-

petitive Product Markets: Welfare Implications", Environmental and Resource Economics, 9,

65-81 (1997).

8. APPENDIX Appendix A

In the …rst stage, …rm j quotes the price of permits (q) that solves the following optimization problem

max

f p

y

c

[y

] + q (E

S) h

[ y

S + E

] g (14) s.t.

p

= p

[E

[q]];

y

= y

[E

[q]];

E

= E

[q];

q 0:

where (i) x

= x

= E

S corresponds to the amount of permits sold (or bought, when x

< 0) by …rm j; and (ii ) a

= y

S + E

corresponds to the amount of polluting emissions abated

by …rm j:

In the second stage, …rm i observes q and chooses the part of polluting emissions it wishes to cover with permits E

by solving the following optimization problem.

max

f p

y

c

[y

] + q ( S E

) h

[ y

E

] g (15) s.t.

p

= p

[E

];

y

= y

[E

];

E

0:

where (i) x

= S E

corresponds to the amount of permits sold (or bought, when x

< 0) by …rm i; and (ii ) a

= y

E

corresponds to the amount of polluting emissions abated by …rm i: The solution to problem (15) de…nes the equilibrium level of permits used for production conditional on permits price, E

[q]: The choice of E

will then determine the amount of permits to buy. The amount of permits traded in equilibrium, conditional on permits price is equal to:

x

[q] = x

[q] = S E

[q]: (16)

Finally, in the third stage, given …rms’choices regarding optimal use of permits for production and the consequent amount of permits bought or sold, …rms strategically compete à la Cournot in

1 2From the market clearing condition,xi = xj: Sincexi= S Ei andxj = (1 )S Ej;we obtain that Ei+Ej=Sand, accordingly,aj= yj Ejcan be written asaj= yj S+Ei:

the output market. Given q and E

[q]; the equilibrium vector y

[E

; q]; y

[E

; q] simultaneously solves the following optimization problems:

max

f p

y

c

[y

] + q ( S E

) h

[ y

E

] g ; (17) s:t:

y

0

max

f p

y

c

[y

] + q (E

S) h

[ y

S + E

] g ; (18) s:t:

y

0

Relying on backward induction techniques, the previous optimization problems can be solved se- quentially starting from the last stage of interaction. In our analysis, we focus exclusively on interior solutions, in which the …rst order conditions lead to y

> 0; E

> 0; and a

> 0. The variable x

can be positive or negative. When …rm k is a net seller of permits, it follows x

> 0: Conversely, when …rm k is a net buyer of permits, x

< 0:

Proof of Lemma 1

The …rst order conditions in (5) correspond to …rms best response functions y

[E

; y

] and y

[E

; y

]: These functions de…ne …rms’ optimal output conditional on the strategy of the rival.

To compute the derivatives

i

and

i

; one must take into account the direct e¤ ect of E

(directly incorporated in …rms’best response functions) but also the indirect e¤ect of E

; which a¤ects …rms’ decisions through the output decision of the rival (the so-called strategic e¤ ect

).

More precisely, the derivatives can be obtained from …rms’best response functions as follows:

@y

[E

]

@E

=

@yi[Ei;yj]

@Ei

+

j

@yj[Ei;yi]

@Ei

1

j

@yj[Ei;yi]

@yi

; (19)

@y

[E

]

@E

=

@yj[Ei;yi]

@Ei

+

i

1

i

@yi[Ei;yj]

@yj

; (20)

where the application of the theorem of the implicit function to the …rst order conditions allows us to obtain the derivatives

i

;

i

;

j

; and

i

.

1 3For a rigorous di¤erentiation between direct e¤ect and strategic e¤ect, see Tirole (1988).

Conditions in system (5) de…ne two equalities:

(E

; y

; y

) = 0 and

(E

; y

; y

) = 0; where:

(

i

(E

; y

; y

) = p

[y

; y

] +

y

c

[y

] h

[a

]

j

(E

; y

; y

) = p

[y

; y

] +

j

y

c

[y

] h

[a

] (21)

By the theorem of the implicit function:

@y

[E

; y

]

@E

=

@ _i(Ei;yi;yj)

@Ei

@ _i(Ei;yi;yj)

@yi

= h

@² i(yi;yj;Ei)

@y²_i

> 0; (22)

with

i

= 2

i

c

h

< 0:

Similarly, the derivative:

@y

[E

; y

]

@y

=

@ _i(Ei;yi;yj)

@yj

@ _i(Ei;yi;yj)

@yi

=

@pi[yi;yj]

@yj

@² i(yi;yj;Ei)

@y²_i

; (23)

which is negative due to the negative denominator (second order conditions) and the negative numerator (given the properties of the demand functions). In the light of the negative sign of the derivative

j

, …rms’decisions are strategic substitutes, which is always the case in the context of quantity competition. In addition, we observe

1 < @y

[E

; y

]

@y

< 0; (24)

given that

>

j

:

The derivative

@y

[E

; y

]

@E

=

@ _j(Ei;yi;yj)

@Ei

@ _j(Ei;yi;yj)

@yj

= h

@² j(yi;yj;Ei)

@y_j²

; (25)

which is negative.

Concerning strategic substitutability of quantity decisions, we obtain:

@y

[E

; y

]

@y

=

@ _j(Ei;yi;yj)

@yi

@ _j(Ei;yi;yj)

@yj

=

@pj[yi;yj]

@yi

@² j(yi;yj;Ei)

@y_j²

; (26)

with 1 <

i

< 0: